Thursday, March 15, 2018

The Pros and Cons of Bucket Strategies

Continuing recent posts updating my past descriptions of retirement strategies, let's look again at time-segmentation (TS) or "bucket" strategies.

The basic implementation of time segmentation strategies sets aside enough cash and short-term bonds to cover the next few years of retirement expenses, let’s say five, then covers the following few (let's say years six through ten) years' expenses with intermediate bonds, and finally allocates any remaining assets to stocks.

Note that this is a markedly different way of allocating assets than is typically used by other strategies that base equity allocation on the largest loss a retiree can stomach in a market downturn and the optimal asset allocation to avoid prematurely depleting a savings portfolio.

Many retirees will find that setting aside five or six years of expenses in a cash fund will be a significant portion of their investable assets so this might be a dramatically different allocation.

Here's an example. A retiree wants to spend $40,000 annually from a $1M portfolio. She invests a little less than $200,000 in cash and short-term bonds (the discount rate is low, especially in today's capital markets, so we can roughly just multiply annual spending by 5) to cover expenses for the next five years. She invests a little less than $200,000 (less because they yield a little more) in intermediate bonds and roughly $600,000 in stocks. It is her estimated future spending that determines her asset allocation of 20% cash, 20% bonds and 60% stocks.

TS strategies don't typically recommend an annual safe spending amount like the $40,000 in this example but this can be estimated by any of the (preferably variable) spending strategies.

This part of the TS strategy is based on matching asset duration[1] to the duration of expenses and is financially sound.

Asset duration, in simplest terms, refers to the recovery period typically needed after a market downturn or interest rate increase. The duration of an expense is essentially the number of years until it is due but expected inflation must be considered, too.

Matching a near-term liability with a long-duration asset like stocks would provide a greater expected return but less confidence that the money would actually be available when needed if stock prices declined. Matching a long-term expense with an intermediate bond would have greater certainty but a lower expected return. "Liability matching" provides the greatest asset return for which the expense can be reliably met and is a key component of TS strategies.

Short-term bonds may have a duration in the neighborhood of three years, intermediate bonds 5-10 years, and stock market duration is measured in decades. This simply means that we can be pretty sure of the value of a 3-year bond in three years but not less, the value of cash next year, and that we probably need to invest in stocks for 7-10 years to be pretty sure our investment won't be looking at a loss.

Planners often recommend that the bonds are set up as a ladder held to maturity to mitigate interest rate risk but many planners simply use bond funds of short and intermediate durations assuming that the results will be "close enough" to those of bond ladders.

The expressed goals of TS strategies are to match expense durations to asset durations, to help retirees better understand the purpose of their different assets, and to weather bear markets without the need to sell stocks at depressed prices and thereby avoid “panic selling” in a market downturn.

Liability-matching is a sound financial policy, while the latter two are primarily psychological benefits. In fact, from a financial perspective, to quote Wade Pfau:
... it must be emphasized that on a theoretical level, income bucketing cannot be a superior investing approach relative to total returns investing.[2]
The reason is that bucketing typically requires a much larger cash and short-term bond allocation than other (total return) strategies. The difference between the returns available from these two assets and what their value might have earned in the stock market is referred to as "cash drag." You simply earn less money if a larger portion of your portfolio is held in cash instead of invested in stocks.

In a paper entitled, “Sustainable Withdrawal Rates: The Historical Evidence on Buffer Zone Strategies[3], authors Walter Woerheide and David Nanigan showed that the drag on portfolio returns from holding large amounts of cash can be significant.

In other words, the comfort of a large cash bucket can come with a heavy cost. According to the authors, the performance drag imposed by a large cash bucket actually leaves the typical portfolio less sustainable. That suggests that TS strategies increase the security of income for the next ten years but do so at the cost of less security of income in the years beyond.

Said differently, the goal of TS strategies is to reduce sequence risk, i.e., to reduce the probability of outliving one's savings, by encouraging the investor to avoid selling stocks at low prices. Woerheide and Nanigan, however, show that this strategy's cash drag is typically greater than the benefit of avoiding selling low and often achieves the opposite, a less sustainable portfolio.



Are bucket strategies easier for retirees to understand or is their explanation simply easier to get away with?
[Tweet this]


The goal of building a cash bucket to weather bear markets conflicts with the goal of maximizing portfolio returns (and thereby increasing portfolio sustainability) and can backfire. The longer the cash bucket the greater the cash drag. The shorter the cash bucket the less likely it is to outlast a bear market.

It has also been argued that TS strategies reduce sequence of returns risk and they probably do when they work, meaning when they outlast the bear. But, Moshe Milevsky showed in “Can Buckets Bail Out a Poor Sequence of Investment Returns?” that this strategy cannot always avoid sequence risk. When a retiree spends all his cash in a market downturn he can be left with an extremely risky all-equity portfolio, possibly before the bear market ends, with the postponed selling having had the unhappy effect of waiting to sell stocks until near the market's bottom.

Technically, bucket strategies are not “floor and upside” strategies but, as I have noted in previous posts, most Americans are eligible for Social Security retirement benefits. Consequently, most Americans have a floor, no matter which strategy they prefer, though it may not be what a retiree would consider an adequate floor — living off Social Security benefits alone isn't pretty.

While floor-and-upside strategies are meant to provide confidence that the retiree will never fall below a certain level of income for a lifetime, bucket strategies attempt to inspire that confidence (not always justified, as Milevsky explained) for only the length of the bond ladder.

I often think of the floor issue by imagining that my upside portfolio has been completely depleted. This is an unlikely scenario to be sure unless one is spending from that portfolio, but it forces me to imagine my circumstances in a potential failure scenario. Using a bucket strategy, I would have no stocks from which to replenish the longest rungs of the bond ladder in that event, so my income beyond this "rolling ladder" is clearly dependent upon equity performance and is not secure.

The stock allocation will decline with age as the short- and intermediate-term buckets slowly come to dominate the portfolio. At some age, the portfolio will contain mostly bonds and cash.

TS strategies recommend spending first from cash, then from bonds, then from equities, but as the Michael Kitces explains[4], that is pretty much what happens when we rebalance a SWR portfolio. Rebalancing results in selling assets that have recently experienced the highest growth. If stock prices have fallen, rebalancing ensures that it is other asset classes that will be sold. With rebalancing, stocks are sold after their prices go up.

Lastly, how many retirees — or planners, for that matter — understand these risks?

It's simple enough to explain to a retiree where the funds are coming from to pay bills for the next several years. But, unless she also understands that buckets can fail, that increasing the cash bucket to avoid failure dilutes her expected portfolio returns, and that income for future years funded by the stock market is still at risk, then this benefit of bucket strategies is not a true understanding but simply a psychological salve.

If that's the case, are bucket strategies easier for retirees to understand or is their explanation simply easier to get away with? The arguments for bucket strategies are not that the strategy itself is easier to understand but simply that it is easier to understand the purpose of their asset allocations.

Planners report that bucket strategies improve the bear-market behavior of their clients and their planners find that quite valuable. There's nothing wrong with that if the retiree understands the cost of this behavior management — the long-term sub-performance of an overly-conservative TS portfolio is likely outweighed by the losses they avoid by not selling low.

It's hard to evaluate that comparison because it is largely dependent upon the retiree's self-control but it seems a steep price to pay for this guardrail, especially compounded over a long retirement.

A strong urge to sell in bear markets could just be a sign of an overly aggressive asset allocation. Finding a more tolerable asset allocation between that and a 20% cash allocation might be a better answer.

No retirement funding strategy is perfect and I think a sub-optimal strategy is better than no strategy. Or, as my friend, Peter is fond of saying, bad breath is better than no breath at all.

Ultimately, I firmly believe that the best retirement plan is the one that lets you sleep at night.



REFERENCES

[1] Efficient Frontier, William Bernstein.

[2] The Yin and Yang of Retirement Income Philosophies, Wade D. Pfau, Jeremy Cooper.

[3] Journal Sustainable Withdrawal Rates: The Historical Evidence on Buffer Zone Strategies, Walter Woerheide and David Nanigan.

[4] Is A Retirement Cash Reserve Bucket Unnecessary?, Michael Kitces.

26 comments:

  1. We are keeping a large cash balance so as to not have any large capital gains in the next few years getting 1.5% in a money market fund. I don't consider it a bucket strategy as our expenses are small compared to investible assets.

    ReplyDelete
  2. Does this make sense? For people planning on earlier retirement,(like me) say 3 to 5 years away from S.S. and Medicare, maybe keeping what you need over the next 3 to 5 years in Cash (including health care costs), mostly in CD's and high interest Money Market accounts or equivalent and keep the rest in your Stock and Bond Portfolio and continue to hopefully let your portfolio grow untouched until you reach your official retirement?

    I know that's a big "it depends" on how much you actually have and if the Cash part you need will not hurt what I call your actual real retirement portfolio plan starting at say 65 - 67.

    ReplyDelete
    Replies
    1. Yes. That's different than holding 5 years of cash for your entire retirement.

      But, there aren't any "high interest" money market accounts these days. You'll need to settle for a piddling/interest money market account.

      Thanks for commenting!

      Delete
  3. Dirk,
    I'm using a modified bucket approach. I have 20% of my portfolio in short/intermediate bonds as my security cushion, representing 5 yrs at 4%. This money is only to cover me during market downturns. The remaining 80% is invested in equities(stocks & funds). So far (I retired in Feb 2011) the strategy is working fine. Apparently I'm one of the lucky ones retiring on the crest of a wave, but my CAGR is at 10.7% after 7 years.

    I NEVER expected things to work out this well, but after 7 years of withdrawals, I have substantially MORE than I started with...

    ReplyDelete
    Replies
    1. That's great news, Doug, but two points.

      First, don't confuse a fortunate time to retire with a good strategy. It's reasonable to assume you would have done even better had you been holding less cash, which you would have with almost any other strategy. The strategy is valuable when things don't go well. If we could depend on the returns of the stock market for the past 7 years, the only strategy retirees would need is "invest it all in stocks." Any reasonable strategy works in a bull market.

      Second, as I often have to remind readers of my blog, don't declare victory too soon. Hardly anyone depletes a portfolio in less than 20 years. See my Death and Ruin post from January 2016.

      Here's wishing you continued great luck!

      Delete
  4. Oh, I heartily agree! That's why I mentioned my LUCKY timing, rather than my stock picking skills. LoL!!

    Everyone is in a different place with different circumstances, so no single approach will work for all. My point was to illustrate some actual results with a two bucket approach, main investments + safety cushion.

    ReplyDelete
    Replies
    1. I understand but my point is that the results that you’re illustrating are solely from a 7-year bull market.

      Regardless, I’d rather be lucky than good and either way you’re off to a great start!

      Cheers. . .

      Delete
  5. As the saying goes:"Some people say that life is strange, but compared to what?" The problem of a reduced overall return are caused by insisting here on, what, in effect, becomes a 60/40 portfolio due to the ten year bond/cd ladder. After all, 10 times 4% = 40% and that probably is an excessive fixed income allocation. Indeed, it is worth noting that Bill Bengen came up with the "4% Rule" using portfolios ranging from 50% to 75% stocks and he concluded that 75% was preferable. Translating that into TS terms, then, 10 years is probably too long and 5-6 years is probably more appropriate. Will it protect against every conceivable downturn? No. Does that mean it is not helpful? No to that as well.

    Why bother with time segmenting rather than simply using a percentage allocation? There are two reasons. The first is practical. By giving the bonds a specific function to perform in the portfolio, you address the issue of what type of bonds and what maturities to use and that can have a huge impact. The second reason is more psychological, but still very powerful. I tend to focus on a five year time horizon and what I tell clients is that while no one can say that a downturn won't last longer than five years, what this does is to give us five years to figure out what is going on and how best to respond to it. So what? Well, I can only tell you that the closest I came to a panicked phone call in the 2008 debacle was a sole client who called and asked me to "go over that five year thing, again." We did and that was that.

    ReplyDelete
    Replies
    1. I largely agree with you, but let's look at your argument in a little greater detail.

      First, the 40% bond allocation is a red herring. The example I chose just turned out that way. As the Kitces post I referenced shows, the bond allocation can end up much higher.

      I don't put a lot of stock in 4% Rule asset allocations because I don't put much stock in fixed-spending strategies, but even if I did, when we talk about a 40% cash/bond allocation we're not typically referring to 20% or more of it in cash. I don't find 40% in bonds unreasonable, at all, but a cash allocation of 5% to 10% is more reasonable. It's misleading to suggest that this is a typical 60/40 portfolio.

      "By giving the bonds a specific function to perform in the portfolio, you address the issue of what type of bonds and what maturities to use and that can have a huge impact." That's true, but bond ladders and liability matching are not exclusive to TS strategies. You can use them in the bond portfolio of every other strategy and that actually wouldn't be a bad idea. You'd just hold less cash. So, liability-matching ladders are not a reason to use a TS strategy.

      When you went over that "five year thing" with your client, did you also explain and did your client understand what would happen if the bear market outlasted her bucket? That the strategy actually makes her long-term portfolio less sustainable? Because if you did, and your client understood the risks and the costs of the strategy and was still happy with it, then I have no argument. A TS strategy is right for her. (I am curious as to how well the "we have five years to figure out a fix" argument goes over.)

      That really only leaves the argument that it made it easier for you to manage your client's behavior and that sounds like more of a benefit to you than to her. With all due respect, I spend my time worrying about retirees and expect you planners to take care of your business.

      Love the comment! I'm always looking for a good argument. :-)

      Cheers!

      Delete
  6. This comment has been removed by a blog administrator.

    ReplyDelete
  7. Dirk, I think the opportunity cost of holding significant cash is just too great.
    There are great websites that will give investors the correlations between stocks, bonds, mutual funds,market indicators, alternative investment funds, etc.
    1) Choose extremely low or even negative correlations to one's equity portion and 2) Check how this investment performed during the recent 10% S&P correction period, 1/28/2018 - 2/08/2018. E.g., SPFPX is an alternative bank loan fund, was actually positive during the correction period, has very low market correlations and at its worst has short-lived and minor dips in its performance graph - it's a fairly straight line, quite dependable - and has performed in the 6% area.

    ReplyDelete
    Replies
    1. I think most academics and economists have come to that conclusion (your first sentence), as well.

      Thanks for commenting!

      Delete
  8. Most articles that address allocation and spending "in retirement" seem to assume that income for retirees is limited to Social Security, pensions, and the "nest egg." What if rental property is continuing to provide enough or nearly enough for the predictable future and the nest egg is likely to be surplus funds. How then to allocate the nest egg?

    ReplyDelete
  9. I am not particularly smart so forgive me if I show a fundamental misunderstanding here. My interest is because I am about to retire, hold a large percentage of cash, in savings and pension funds, and was thinking along TS lines, almost exactly as you describe! What puzzles me is in a TR strategy, say 60/40 equity to bonds, what do I sell to get the withdrawal and how does this mesh with rebalancing?

    Do I look at the prevailing allocation and sell to rebalance and as I get older do I put the equity/bond allocation on a glide path towards more bonds?

    Thanks for your blogs, especially the ones like this one that challenge my thinking.

    Paul

    ReplyDelete
    Replies
    1. Paul, despite the self-deprecation, you ask smart questions.

      If you decide on a TS strategy, most of these questions will be answered for you. TS takes a different asset allocation approach than most strategies. Your asset allocation will be largely decided by how much you plan to spend and when. Other strategies base asset allocation on your short-term risk tolerance and on maximizing portfolio sustainability (not outliving your savings). With TS, you are likely to end up with a very conservative asset allocation that becomes more aggressive (riskier) during a long bear market.

      First, ignore glide paths. There is no reason to decide at age 65 what your asset allocation will be at age 85, for example, when you have no idea what your financial situation will be then. As Wade Pfau once told me, "Glide paths are for managers of target-date mutual funds or what we would recommend for individuals if we knew nothing about them except their age. A custom asset allocation is always best."

      Said differently, decide on your asset allocation at age 80 (or whatever) when you get there. There's no advantage to deciding sooner.

      The asset allocation is set for you by the TS strategy. You don't start out with the notion of a 60/40 allocation. You invest the next 5 years' expected spending in cash, the following five years' expected spending in intermediate-term bonds and the rest in stocks and you live with whatever equity allocation that gives you.

      It's important to add that a 60% equity/40% bond portfolio with 5% of the 40% invested in cash (a typical total returns allocation) is quite different than a 60/40 portfolio with 20% or more of the 40% held in cash, as a TS allocation might require. Don't be confused by calling both "60/40" allocations.

      In a long bear market your allocation will become progressively more aggressive as you spend cash and perhaps sell bonds. You live with that, too.

      What do you sell? You spend from the cash bucket first, the bond bucket second and equities last. You replenish cash and bonds in good market times by selling stocks. During a bear market, you postpone replenishing the buckets as long as you can because the goal is to avoid selling stocks at low prices.

      This means that rebalancing is also taken care of for you, though not exactly in the best way. In contrast, rebalancing a total returns portfolio sells stocks when stock prices have gone up.

      So, to summarize. The TS strategy will determine your asset allocation for you and rebalance for you, though not always in the most efficient way. Ignore glide paths. Spend first from the cash bucket, then from bonds and last from stocks. Don't sell stocks during a bear market if you can spend from cash or bonds.

      I suspect your confusion stems from the fact that TS strategies do things quite differently than total returns strategies. Many planners argue that TS strategies make it easier to console clients in a market downturn. That's a different argument than TS strategies are easier to understand, wouldn't you say?

      Thanks for the question!

      Delete
  10. Good question. When building your retirement plan, simply make some prediction about how much rental income you expect and for how long. Like any other investment, real estate has risks. You may reach an age, for example when you no longer want the hassle. A friend in the real estate business declared bankruptcy in 2008. There are risks to everything.

    To the extent that you can pay your bills from rental income and Social Security, you don't need to touch that nest egg. Since you are less dependent on portfolio spending, you can invest your portfolio more aggressively (a higher equity allocation) than you otherwise would and you won't need to worry as much about its volatility.

    For example, if most of your bills are otherwise covered, you might be able to invest 80% of your portfolio in equities instead of 40% to 60%.

    I wouldn't invest any portfolio funds in real estates, though. You probably already have more than enough.

    Thanks for writing.

    ReplyDelete
  11. It strikes me that when it comes time to withdraw from and re-balance a bucket strategy that re-balancing is based on market timing, i.e. knowing when the equity market is depressed and ready for a regression to the mean. Market timing is widely regarded by the professional community as a fool's errand.

    For a practical approach to withdrawing from retirement savings, called 3-PEAT, see https://www.onefpa.org/journal/Pages/AUG17-A-3-Step-Procedure-for-Computing-Sustainable-Retirement-Savings-Withdrawals.aspx

    The retiree who practices 3-PEAT, computes her retirement plan for the current year and executes her planned withdrawals during the year. Next year she repeats the process with new data based on updated account balances, changes in her personal situation, and changed market conditions. Her plan includes not only asset allocations across the three types of savings accounts (IRA, RothIRA, Taxable), but also Social Security Benefits, Pensions, annuities, sale of illiquid assets (her home), and her glide path strategy.

    The 3-PEAT simulator, rather than trying to predict the future, assesses the performance of her plan for selected historical market periods. Plan performance is measured, not by plan success or failure, but by what is the smallest, after-tax, real, disposable income for any year in the historical retirement period. 3-PEAT does not fail in that savings are never fully depleted. There are always savings available to take advantage of any subsequent bull market, after a sever downturn. While savings will not fail the retiree may be alarmed by the amount of money available spending during a particularly lean year.

    There are Bogleheads who have been practicing 3-PEAT principles, without calling it that, for some years now. The 3-PEAT simulator confirms that what they are doing is safe and practical.

    It should be noted that while there are quantitative studies identifying the better glide path strategies, the bucket literature is not quantitative, but mostly expert conjecture.


    ReplyDelete
    Replies
    1. Jim, I think I would agree that TS rebalancing is essentially market timing, since the retiree has to guess when the bear is over and it's time to replenish buckets. Though, rebalancing may also be forced, and at an inopportune time, by a long bear market.

      The method you suggest, however, is based on assuming that historical returns repeat. It assumes that the market mean-reverts and even if it does (I suspect it does) we don't know how long it will take or if the underlying market return changes over time. (In other words, to what mean does it revert?) That's a few too many assumptions for me.

      As I mentioned above, I'm not a fan of glide paths for individual households. I see no rational reason for an individual household to select an asset allocation in advance or, as Wade Pfau says, a custom allocation is always best.

      Thanks for the thought-provoking comments.

      Delete
  12. I have been reading your posts on retirement withdrawls and it has been a very interesting and informative read. I have also researched other sources of information. I will be retiring soon and have decided to use a 5% withdrawl rate not linked to inflation, simple and with inbuilt flexibility.

    ReplyDelete
    Replies
    1. That is a very dangerous strategy for multiple reasons. I recommend that you don't.

      Delete
  13. If you run some simulations at the Firecalc site to determine a SWR based on asset allocation the results do seem to support this article. For a 30 year period with 50% in US stocks the SWR is about 3.75% if fixed income is held in 5 year treasuries. It drops slightly to 3.73% is the fixed income is held as "commercial paper", which is as close as I can find to "cash". While this is not terribly significant, even this very basic simulation demonstrates that there can be a drag with cash. As an aside, the SWR using 30 year treasuries drops to about 3.66% (From these three data points (!!), one could conclude that Bengen's original work in 1994 using 50% US stocks and 50% 5 year treasuries might be the optimum).

    ReplyDelete
    Replies
    1. Thanks for the comment. A lot of recent research has concluded precisely that. Wade Pfau currently believes the safe withdrawal rate is probably closer to 3%.

      Delete
    2. Here is one other example from another one of the common web tools - Portfolio Visualizer. This one is more anecdotal, since it only looks at a single time period, but is illustrative. Using the tool's full data set (1972 to present) and arbitrarily choosing a 4% withdrawal rate to run the simulation the results are as follows:

      For portfolios with 50% large cap US stocks starting in January 1972, the portfolio holding the balance in cash will be depleted by 2013, while the portfolio holding intermediate bonds will have retained nearly 40% of its inflation-adjusted original value. Further, by 2018, the portfolio with bonds is still going at about 30% of its inflation-adjusted original value. while, again, this is VERY anecdotal, it is one illustration of a period where cash was a huge drag.

      As an aside, I really enjoy Pfau's papers, and am in no position to contest his work, but I personally have settled on a SWR of 3.6% for planning of my retirement based on simulations done with my portfolio and expected duration of a conservative 35 years. I would only say that if Pfau is correct at SWR of 3% or slightly less, then things like inflation adjusted SPIAs may be attractive.

      Delete
    3. Anecdotes are interesting but a dangerous way to plan anything. I believe Wade's results suggest 3% or a little more as opposed to 3% or less and I think he would probably find your 3.6% choice quite reasonable.

      The fact remains that no one knows how the stock market will perform over the next three decades. Good luck and thanks for sharing!

      Delete
  14. Thanks for finally writing about >"The Pros and Cons of Bucket Strategies" <Loved it!

    ReplyDelete
  15. I am a few years from retirement and we will likely have very good Social Security income with some supplemental pension income. Our "floor" should be above the median household income which gives us an appreciable amount of safe income.

    So my current perspective on buckets is to build a bucket of cash and short-bonds in my early 60s to provide secure funding to replace the "lost" Social Security income from delaying my SS from 66 yrs 10 months to 70. At 70, my SS will likely be at or close to the maximum available, and the need for a large cash bucket would go away as our floor income should be relatively secure. We have significant pre-tax savings that should generate as much or more as the SS and pension income at something like 5% per year withdrawals.

    After age 70, I assume the buckets would be much smaller, focusing on providing the next year or so withdrawal in cash so that a market plunge doesn't suddenly disrupt your plans in the next year. Additional cash or term-matching bonds could be set aside for specific large expected expenditures in the next 5 years if you don't want them subjected to market risk.

    We will likely do some conversions to Roth IRAs or start contributing to a Roth 401k if my employer starts that in the next year or so. That would become a non-cash bucket with a more aggressive investment profile than the pre-tax account until mid-80s at which time it would drop back to 50% or less equities. The goal for the Roth would be to not tap it until the high RMDs in late 80s-90s start to deplete the pre-tax account, especially in a bear market.

    ReplyDelete